Re: [Intel-xe] [PATCH] drm/xe/pm: WARN_ON() on unbounded pcie parent bridge device

[Rodrigo Vivi <rodrigo.vivi@kernel.org>]

On Tue, May 09, 2023 at 03:20:25AM +0000, Gupta, Anshuman wrote:
> 
> 
> > -----Original Message-----
> > From: Vivi, Rodrigo <rodrigo.vivi@intel.com>
> > Sent: Monday, May 8, 2023 10:45 PM
> > To: Gupta, Anshuman <anshuman.gupta@intel.com>
> > Cc: intel-xe@lists.freedesktop.org
> > Subject: Re: [Intel-xe] [PATCH] drm/xe/pm: WARN_ON() on unbounded pcie
> > parent bridge device
> > 
> > On Mon, May 08, 2023 at 02:47:34PM +0530, Anshuman Gupta wrote:
> > > Intel Discrete GFX cards gfx may have multiple PCIe endpoints, they
> > > connects to root port via pcie upstream switch port(USP) and virtual
> > > pcie switch port(VSP), sometimes VSP pcie devices doesn't bind to
> > > pcieport driver. Without pcieport driver pcie PM comes without any
> > > warranty and with unbounded VSP gfx card won't transition to low power
> > > pcie device and link state therefore assert drm_WARN_ON() on
> > unbounded
> > > VSP.
> > >
> > > Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
> > > Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
> > > ---
> > >  drivers/gpu/drm/xe/xe_pci.c | 13 +++++++++++++
> > >  1 file changed, 13 insertions(+)
> > >
> > > diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
> > > index 19ead642442e..dc0d5cb4950b 100644
> > > --- a/drivers/gpu/drm/xe/xe_pci.c
> > > +++ b/drivers/gpu/drm/xe/xe_pci.c
> > > @@ -667,12 +667,24 @@ static void xe_pci_shutdown(struct pci_dev
> > *pdev)
> > >  	xe_device_shutdown(pdev_to_xe_device(pdev));
> > >  }
> > >
> > > +static void xe_assert_on_unbounded_bridge(struct pci_dev *pdev) {
> > > +	struct pci_dev *bridge = pci_upstream_bridge(pdev);
> > > +	struct xe_device *xe = pdev_to_xe_device(pdev);
> > > +
> > > +	if (!bridge)
> > > +		return;
> > > +
> > > +	drm_WARN_ON(&xe->drm, !bridge->driver); }
> > > +
> > >  #ifdef CONFIG_PM_SLEEP
> > >  static int xe_pci_suspend(struct device *dev)  {
> > >  	struct pci_dev *pdev = to_pci_dev(dev);
> > >  	int err;
> > >
> > > +	xe_assert_on_unbounded_bridge(pdev);
> > 
> > besides warning, shouldn't we block the suspend at all?
> > returning from this point...
> If that is desired then, we shall check this in xe probe and initializes the PM OPS with NULL rather then failing suspend ?

great idea!

> Br,
> Anshuman Gupta. 
> > 
> > >  	err = xe_pm_suspend(pdev_to_xe_device(pdev));
> > >  	if (err)
> > >  		return err;
> > > @@ -718,6 +730,7 @@ static int xe_pci_runtime_suspend(struct device
> > *dev)
> > >  	struct xe_device *xe = pdev_to_xe_device(pdev);
> > >  	int err;
> > >
> > > +	xe_assert_on_unbounded_bridge(pdev);
> > >  	err = xe_pm_runtime_suspend(xe);
> > >  	if (err)
> > >  		return err;
> > > --
> > > 2.38.0
> > >